[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: rawhide build problems

On Mon, 2014-01-20 at 08:06 -0800, Brian C. Lane wrote:
> On Sat, Jan 18, 2014 at 09:21:25AM -0800, Adam Williamson wrote:
> > 
> > I guess if I were you I'd just use a mock environment or a VM or a
> > Docker instance (OO LOOK SHINY) to do this?
> Thanks for tracking those down. There may be some dependency problems
> with gtk3 then, since I installed this using yum --enablerepo=rawhide
> update gtk3 gtk3-devel and it only pulled in 3 packages
> (gtk3-immodule-xim is the 3rd).

I think you may be right. At the package level, gtk3-devel depends on a
couple of pkgconfig things provided by glib2, but they aren't
specifically versioned:


Ditto Wayland - it depends on:

pkgconfig(wayland-client) >= 1.2.0

It looks rather like, in theory, it should require something like:

pkgconfig(gio-2.0) >= 2.39
pkgconfig(wayland-client) >= 1.3.90

But this should probably be fixed at the upstream level, as it looks
like these deps are either manually or automatically generated from
upstream. Upstream is https://git.gnome.org/browse/gtk
+/tree/configure.ac , and indeed it looks like the pkg-config dep on
gio-2.0 is unversioned and the dep on wayland-client is versioned 1.2.0.
Both of those should probably be bumped.

CCing mclasen. mclasen, see
https://www.redhat.com/archives/anaconda-devel-list/2014-January/msg00020.html and  https://www.redhat.com/archives/anaconda-devel-list/2014-January/msg00021.html for the background here, why these changes I think would be correct for gtk+ upstream. I can submit a patch if you'd like.

> So for now I guess I'll see if I can setup a minimal working rawhide to
> use for building. whee.

That's probably a good idea in general, because it's been my experience
that you can't really rely on this kind of dependency being correctly
stated. It is quite a large burden on packagers to maintain versioned
dependencies to such a degree as to permit mixing of packages between
releases even where it's viable (i.e. what you're doing), and it
generally doesn't really happen.

It does seem like this specific instance is one it would be a good idea
for GNOME/GTK folks to fix upstream, though.
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]